Abstract:
Clustering is the process of partitioning or
grouping a given set of patterns into disjoint
clusters. Clustering web usage data requires
developing specialized techniques based on the
web log data. The methodology has to improve the
data preprocessing as well as the quality of the
clusters. Traditional K-means algorithm is widely
used clustering algorithm with wide range of
application. Since K-means algorithm has some
disadvantages. The proposed system intends to
implement an enhanced K-means clustering
algorithm to get more accurate and effective
cluster results. The algorithm will carry out in
two steps which include finding initial centroids
and clustering stage. The testing of the system
uses web log data from NASA web usage data for
clustering. The performance and accuracy of the
system are also measured.
